[New] SPS: Introduction to Product Quality Score

image.png
Quick Navigation: Where should you start?
  • Product Quality Score calculation steps → How is product quality score calculated?
  • Whether appeals are supported → How to Appeal
  • How to check Product Quality Score? → Feature Guide

What Is the Product Quality Score?

The Product Quality Score is a rating system that measures product experience. It refines quality assessment from the shop level down to the single-product level to pinpoint potential experience problems.
  • Measures product performance: Uses a 0–5 score for the product's quality; a higher score means a better user experience.
❗️Key change: product score calculation moves from the "shop dimension" to the "product dimension"
  • Old version (Product Satisfaction): Aggregates all shop orders to calculate NRR and NBFR, calculating one shop-level score.
  • New version (Product Quality Score): Scores each product first, then aggregates by "product score × the product's order share" into the shop-level Product Satisfaction
    • Special note: Sellers whose shop calculation category is Pre-owned will continue to be calculated using the old version’s shop-level overall calculation method. For details, see FAQ: Why does Product Satisfaction for the Pre-owned category still use shop-level evaluation?
The following section explains the calculation steps for the quality score of an individual product.
image.png

How to Calculate Product Quality Score?

Calculation Formula

Product Quality Score = NRR score × weight + NBFR score × weight
Positive Feedback Rate doesn't enter the score calculation and only serves as a high-score threshold: to reach 4.5 or above, a product must meet the Positive Feedback Rate standard; if it fails, the product score is capped at 4.4.
Shop-level Product Satisfaction = each product's quality score in the shop × each product's order-share weight
Metric
Time window
Definition
Metric type
Negative Review Rate (NRR)
90D
Of a product's delivered orders in the past 90 days, the share that initially received a 1-star or 2-star product review.
  • [Updated on 10/16] The system no longer uses the first review for an order. Instead, it uses the latest review status within 15 days after order delivery.
  • Excludes negative reviews from FBT or 4PL logistics issues.
Scoring metric
Non-Buyer Fault Return & Refund Rate (NBFR)
90D
Of a product's delivered, non-cancelled orders in the past 90 days, the share where a seller- or logistics-partner-caused problem (e.g., package damage) led the customer to request a refund or return.
  • Only the first return/refund request counts.
  • [Updated on 10/26] Excludes:
  1. Orders related to platform-fault logistics issues
  2. Orders determined through arbitration as non-seller-fault
  3. Orders where the after-sales request was canceled after being initiated
Scoring metric
Positive Feedback Rate
90D
Share of a product's 4-star or 5-star reviews among all its reviews in the past 90 days.
High-score threshold

#Step 1: Meet the Scoring Threshold

A product needs ≥ 5 cumulative valid orders in the past 90 days (excluding fake, sample, or gift orders).
Even without a SPS, a product with ≥ 5 cumulative valid orders in the past 90 days will receive a product quality score.
Cumulative valid orders in the past 90 days
Handling
Notes
< 5 orders
No score calculated
Too few valid orders; no score for now.
≥ 5 orders
Calculate product score normally
Meets the calculation threshold; enters scoring.
Special Note:
For the Pre-owned category, each product (PID) typically has only one order. Since the sample size at the individual product level is limited, the result may be unstable or less reliable. To evaluate sellers in this category more fairly:
  1. Sellers whose shop calculation category is Pre-owned will be evaluated at the shop level overall.
  2. [Updated on 10/16] For Pre-owned products under shops whose calculation category is not Pre-owned:
    1. One product, one order: If NRR/NBFR = 0%, the default score is 4.4; if NRR/NBFR = 100%, the default score is 2.5.
    2. Non-one-product-one-order cases: Scored as usual.

#Step 2: Calculate Metrics

  • Negative Review Rate (NRR) = number of product orders with 1–2 star reviews / number of the product's delivered orders in the past 90 days
  • Non-Buyer Fault Return & Refund Rate (NBFR) = number of product orders returned and refunded due to seller or logistics reasons / number of the product's delivered orders in the past 90 days
Metric calculation example: taking Negative Review Rate (NRR) as an example.
A product had 200 delivered orders in the past 90 days and received 10 negative reviews.
NRR = 10 / 200 = 5%
Small order-volume protection
  • Once formal calculation starts, NRR and NBFR are each checked for small order-volume protection by delivered orders (denomintor) and problem orders (numerator). This gives low-volume products a score floor, so a small sample doesn't drag the score down:
    1. 5 ≤ Denominator ≤ 10 orders: default score = 3.5
    2. 10 < denominator ≤ 30 orders, and numerator ≤ 1 order: the metric is calculated normally but applies to a 3.5-point floor. If the actual score is below 3.5, count it as 3.5; Otherwise, use the actual score.

#Step 3: Convert Metrics to Scores

Metrics use a 0–5 point rating system, with score standards set from 1 to 4 points.
  • Metric scores are converted based on the product category.
  • Once a metric meets a score-tier standard, it earns at least that score; better performance leads to a higher score.
image.png
💡 Product Scoring Track Note:
  • This is not based on the shop’s calculation category. It is determined by the product’s actual category and price band.
  • The third-level category is used by default. If there are too few products in that category, the second- or first-level category is referenced.
Score conversion example: taking Negative Review Rate (NRR) as an example.
Product A and Product B belong to Socks, and price band is high. In this track, NRR corresponding to the 3-point tier is 2%, and NRR corresponding to the 4-point tier is 1.5%. NRRs of Product A and Product B over the past 90 days are 1.5% and 1.6%, respectively.
  • Thus, Product A’s NRR metric score is 4 points, while Product B’s NRR metric score falls within the 3–4 point range.
  • Based on Product B’s NRR ranking within the Socks-high, Product B’s NRR metric score is calculated as 3.2 points.

#Step 4: Derive the Product Quality Score

  1. Initial score = NRR score × metric weight + NBFR score × metric weight
Metric weight allocation varies by scoring track.
  1. If the initial score ≥ 4.5, check the high-score threshold
High-score threshold
  • Positive Feedback Rate: ≥ 90%
image.png
Rules for high-score thresholdimage.png
Initial score
High-score threshold check
Final score
< 4.5
No high-score threshold check
Take the initial score
≥ 4.5
Positive Feedback Rate meets the standard
Take the initial score
≥ 4.5
Positive Feedback Rate does not meet the standard
Capped at a maximum of 4.4
Full example:
A product belongs to Womenswear, and price band is high. In the past 90 days, it had 200 delivered orders, including 2 return & refund orders and 5 negative reviews. Assume that, in this track, the 4-point standard is 2.5% for NRR and 1% for NBFR, with each metric weighted at 50%.
  1. Metric calculation:
    • NRR = 5 / 200 = 2.5%
    • NBFR = 2 / 200 = 1%
  2. Metric scores: Both NRR and NBFR meet the 4-point standard for the product track, so both metric scores are 4.0.
  3. Initial product score: Based on the 50% weight for each metric, product quality score = 4.0 × 50% + 4.0 × 50% = 4.0
  4. Threshold check: The initial product score is < 4.5, so the high-score threshold check is not triggered → final product quality score: 4.0

How to appeal?

  • SPS plays an important role in shop exposure, buyer trust, and overall business success, so maintaining a high SPS is critical. However, we understand that SPS drops are sometimes caused by factors beyond the seller's control, such as buyer mistakes, logistics issues, or system errors. The SPS Appeal feature allows sellers to directly initiate appeals and take action on SPS metrics they believe have been incorrectly evaluated.
  • The appealable scenarios and reasons for each SPS metric are as follows:
Metrics
Scenario (Appeal Reason)
NBFR
Non-Buyer Fault Return & Refund Rate (NBFR)
Inaccurate Buyer Selected Reason
Perception Issues
Malicious Return and Refund
Logistics Issues
TikTok System Issue
Negative Review Rate (NRR)
Malicious Reviews
Creator False Promotion
Logistics Issue
IM Dissatisfaction Rate
TikTok System Issue
Uncontrollable Event/Unforeseen incident
Malicious Buyers
After-Sales Handling Time (AHT)
Malicious Buyers
TikTok System Issue
Uncontrollable Event/Unforeseen incident
Seller Fault Cancellation Rate (SFCR)
Malicious Buyers
TikTok System Issue
Uncontrollable Event/Unforeseen incident
Dispatch TimeWeather or Natural DisastersTechnical Issues with TikTok Shop Seller CenterMissing scans by TikTok Shipping - Standard Delivery
Transit TimeWeather or Natural DisastersTechnical Issues with TikTok Shop Seller CenterOrder IDs are Lost or being Returned to Shipper

Feature Guide: How to Check the Product Quality Score?

Step
Module & Features
Product UI & Details
Step 1:
Understand overall performance
[Shop Product Satisfaction Overview]
  1. Current Product Quality Score performance
Unlocked and pending benefits[Shop Product Satisfaction Summary Formula]Each product's score in the shop is aggregated into shop-level Product Satisfaction by order share.
New version adds traffic benefits at the product level: traffic incentives are set for each product based on its product quality score.
image
Step 2:
Identify low-score products
[Product List]
Lists each product in the shop from highest to lowest sales.
A prioritize product with high sales but Poor performance or Needs Improvement, as it may be the key factor lowering your shop-level Product Satisfaction.
  • Click <↑↓Sort by> to sort this column.
  • Orders are sorted by default from highest to lowest based on the selected metric’s order share.
    image
Step 3:
View product score details
[Individual Product Quality Score Overview]
  1. Basic product information and current score performance
Unlocked and pending benefit tiers
Click <View Details> to the detail page of Individual Product
↓
image
[Score Calculation Formula]
Helps sellers understand how an individual product's score is calculated.
Due to the high-score threshold check, an individual product’s score calculation may fall into one of the following three cases:
  1. Preliminary score < 4.5; the high-score threshold check is not triggered.
    image
  2. Preliminary score ≥ 4.5 and passed the high-score threshold check.
    image
  3. Preliminary score ≥ 4.5, but failed the high-score threshold check.
    image
Still have questions about the score calculation? Click <learn more> to view:
[Detailed Calculation Rules]
  1. High-score threshold check rules
  2. Product scoring threshold and small order-volume protection rules
image
Click <learn more> to go to the detailed calculation rules.
↓
image
Step 4:
Identify problem metrics
[Metric Card]
  1. Metric value and score for the current period
Performance level: benchmarked against similar sellers[Metric Analysis]
Track metric trends and respond promptly to any changes. Quickly compare the category benchmark against top sellers.
image
If you have any questions about metric definitions or calculations, focus on this module❗️[Metric Details]Click the button in a metric card to view:
  1. Metric definition
  2. How the metric is converted into a metric score
Metric calculation formula
image
Click the button in the metric card to view metric details.
↓
image
[High-Score Threshold]
Check whether the positive metric meets the high-score threshold.
[Metric Details]Click the button to learn more about the positive metric.
image
Click the button in the metric card to view metric details.
↓
image
Step 5:
Locate affected orders
[Affected orders Details]
View order details within the shop.
Taking NRR as an example, this module shows all orders with 1- and 2-star reviews, along with the related issue type and severity level.
image
image

FAQ

Why do my products have scores, but my shop doesn’t?

Product-level and shop-level scoring thresholds are different.
  • Product level: A product can receive a score once it has 5+ delivered orders in the past 90 days.
  • Shop level: A shop receives an SPS only after it has 30+ delivered orders in the past 90 days.
So some products may already have scores while the shop has not yet met the SPS threshold.
Excludes fake orders, sample orders, and gift orders.

Can a product be excluded from scoring after it is delisted or deleted from the shop?

No. All products and their valid orders within the statistical period are counted, regardless of product status.

Why do two products in the same category have different metric weights and scoring standards?

When setting metric weights and scoring standards for product-related dimensions, both category and price band are taken into account. Each product is assigned to a scoring track based on its category and corresponding price band, then compared and ranked against products in the same track to determine its metric score.
[Updated on 10/16] A 90-day average price logic has been added when determining price bands to reduce the impact of short-term price fluctuations on price band assignment.

Why do Pre-owned shops and products use a special scoring logic?

In the new version, Product Satisfaction is calculated at the product level first, then weighted into the shop score by each product’s order share. A product needs at least 5 valid orders in the past 90 days to be scored.
For Pre-owned, one product ID usually corresponds to only one order, so product-level samples are too small and may lead to unstable results. To ensure fairer evaluation:
  1. Sellers whose shop calculation category is Pre-owned — meaning they mainly sell Pre-owned products — will be evaluated at the overall shop level, with a 90-day window for NRR and NBFR.
  2. [Updated on 10/16] For Pre-owned products sold by shops whose calculation category is not Pre-owned:
    • One-product-one-order: default score is 4.4 when NRR/NBFR = 0%, and 2.5 when NRR/NBFR = 100%.
    • Other cases: normal scoring applies.